/v1/modelsList models
Returns provider-scoped ids (e.g. openai/gpt-4o) you pass as model in chat completions, plus tariff metadata. Browse cards and filters on the public catalogue.
Request examples
…cURL:
…Sample response JSON
…/v1/models/codingCoding models
Public endpoint (no API key required). Returns visible, routable catalogue models tagged coding, same OpenAI-compatible envelope as GET /v1/models.
Request examples
…cURL:
…Sample response JSON
…Model object
Each entry is an OpenAI Model (id, object, created, owned_by) plus RodiumAI fields under rodiumai_*, which the OpenAI SDKs keep in model_extra (Python) or as extra properties (JavaScript).
| Field | Meaning |
|---|---|
| id | The value to pass as model, e.g. openai/gpt-4o |
| rodiumai_provider | slug, name, sort_order of the model's brand |
| rodiumai_display_name, rodiumai_description | Human-readable name and description |
| rodiumai_pricing | pricing_unit, input_per_1m, output_per_1m, cached_input_per_1m, per_image in RODI (decimal strings or null), currency |
| rodiumai_capabilities | context_window, max_output_tokens, input_modalities, output_modalities, supports_streaming, supports_tools, supports_vision, supports_json_mode, supports_reasoning |
| rodiumai_tier | speed (fast, balanced, slow) and quality (basic, standard, pro, max) used by smart profiles |
| rodiumai_status, rodiumai_unavailable_message | available or unavailable, with the reason when unavailable |
| rodiumai_rate_limits | Per-key rpm, tpm, rpd for this model (null = default applies), see Rate limits |
| rodiumai_sort_order | Display order used on rodiumai.io |
| rodiumai_kind, rodiumai_profile | Only on virtual entries: smart_router or smart_profile, and the profile name |
- The list starts with the virtual smart entries (
rodiumai/smart,rodium/*), then catalogue models in site order.served_atis the Unix time of the response. - The list is public. With
Authorization: Bearer, it only contains the models that credential may call.GET /v1/models/codingis public and never filtered. - The catalogue is cached for about 60 seconds.
GET /v1/models/{id}
Returns one model object. The id keeps its slash in the path: /v1/models/openai/gpt-4o. Smart aliases resolve too (/v1/models/rodiumai/smart, /v1/models/fast). Unknown ids, and with a key ids outside its allowed models, return 404 model_not_found.
…Response
…